Sam Altman says your laptop shouldn’t run AI. In healthcare, where the data lives matters.

His point is about utilisation during a memory shortage: “You want maximum utilization in the cloud if you’re trying to deal with this very supply constrained environment.” 1 DRAM is the working memory computers use to run software. In a data center it serves many users through the day; in a laptop it often sits idle 2.

Yes, AI models can run locally, and Altman’s own framing assumes it: people “want local inference” and expect open-source models to run on their own computers 1. His argument is that, while memory is scarce, this is a poor use of it 1.

Critics answer that efficiency is not the only goal. One response, a blog post by the software company Koru, put it this way: “But efficiency isn’t the only thing people are optimizing for. For private notes, control matters more.” 3 For a person or a company, that can mean deciding which model sees their information, where it runs, and whether they need a connection at all 3.

In the US, HHS guidance says a HIPAA covered entity may use a cloud service to store or process electronic protected health information:

Yes, provided the covered entity or business associate enters into a HIPAA-compliant business associate contract or agreement (BAA) with the CSP

HHS Office for Civil Rights 4

The same guidance says the type of cloud setup can change the risk analysis, and that a cloud provider storing only encrypted data is still a business associate 4. HHS is also explicit: “OCR does not endorse, certify, or recommend specific technology or products.” 4 In the EU, data concerning health is one of the special categories of personal data under Article 9 of the GDPR 5.